Historical Associations of Molecular Measurements of Escherichia coliand Enterococci to Anthropogenic Activities and Climate Variables inFreshwater Sediment CoresYolanda M. Brooks,*,† Melissa M. Baustian,‡,§ Mark Baskaran,∥ Nathaniel E. Ostrom,⊥

and Joan B. Rose†,#

†Department of Microbiology and Molecular Genetics, 480 Wilson Road, Room 13, Michigan State University, East Lansing,Michigan 48824, United States‡Center for Water Sciences, Michigan State University, 288 Farm Lane, Room 203, Michigan State University, East Lansing, Michigan48824 United States§The Water Institute of the Gulf, 301 North Main Street, Suite 2000, Baton Rouge, Louisiana 70825, United States∥Department of Geology, 0224 Old Main, Wayne State University, Detroit, Michigan 48202, United States⊥Department of Integrative Biology, 288 Farm Lane, Room 203, Michigan State University, East Lansing, Michigan 48824, UnitedStates#Department of Fisheries and Wildlife, 480 Wilson Rd Rm 13, Michigan State University, East Lansing, Michigan 48824, UnitedStates

*S Supporting Information

ABSTRACT: This study investigated the long-term associa-tions of anthropogenic (sedimentary P, C, and N concen-trations, and human population in the watershed), and climaticvariables (air temperature, and river discharge) with Escherichiacoli uidA and enterococci 23S rRNA concentrations in sedimentcores from Anchor Bay (AB) in Lake St. Clair, and near themouth of the Clinton River (CR), Michigan. Calendar year wasestimated from vertical abundances of 137Cs. The AB and CRcores spanned c.1760−2012 and c.1895−2012, respectively.There were steady state concentrations of enterococci in ABduring c.1760−c.1860 and c.1910−c.2003 at ∼0.1 × 105 and∼2.0 × 105 cell equivalents (CE) per g-dry wt, respectively.Enterococci concentrations in CR increased toward presentday, and ranged from ∼0.03 × 105 to 9.9 × 105 CE/g-dry wt. The E. coli concentrations in CR and AB increased toward presentday, and ranged from 0.14 × 107 to 1.7 × 107 CE/g-dry wt, and 1.8 × 106 to 8.5 × 106 CE/g-dry wt, respectively. Enterococci wasassociated with population and river discharge, while E. coli was associated with population, air temperature, and N and Cconcentrations (p < 0.05). Sediments retain records of the abundance of fecal indicator bacteria, and offer a way to evaluateresponses to increased population, nutrient loading, and environmental policies.

■ INTRODUCTION

The United States of America has a fragmented history of waterquality standards. Culturable total coliforms were first regulatedin drinking water in 1914, and were used to measure ambientwater quality until the 1950s.1 By the 1960s, fecal coliformswere used to monitor wastewater discharges, and then to guidethe safety of recreational waters.2 Since 1986, culturableEscherichia coli (EC) and/or enterococci (ENT) have beenthe indicators of choice for monitoring recreational waters.3

Recently, quantitative polymerase chain reaction (qPCR)measurements of ENT 23S rDNA were recommended by theU.S. Environmental Protection Agency (USEPA) to monitorrecreational water quality.4

Current methods primarily measure fecal indicator bacteria(FIB) in the water column. However, qPCR measurements ofcell equivalents (CE) of ENT and EC and their culturableconcentrations revealed higher concentrations in sediments andbenthic sand, respectively, than the water column.5,6 In the top11 cm of sediment and sand cores obtained from LakeSuperior, there was a ∼ 5-log increase of CE of ENT than theirrespective culturable forming units (CFUs).5 Further, it appearsthat culturable ENT and EC stabilized (<1-log reduction) in

Received: March 18, 2016Revised: June 1, 2016Accepted: June 6, 2016Published: June 20, 2016

Article

pubs.acs.org/est

© 2016 American Chemical Society 6902 DOI: 10.1021/acs.est.6b01372Environ. Sci. Technol. 2016, 50, 6902−6911

sediment microcosms stored at 10 °C for 60 days, indicatingthat sediments have the potential to be long-term reservoirs ofFIB,7 and can negatively impact water quality by sedimentresuspension.8

Specific linkages between water quality trends over large timescales to climate change and human activities continue to bechallenging. An approach which measures FIB in sedimentcores could be used to evaluate water quality changes over largetime scales. For example, a previous study observedconcentrations of culturable EC and ENT in sedimentsdeposited up to 245 ± 45 years ago, and their observedconcentrations increased when eutrophic conditions wereobserved in Switzerland.9 The purpose of our study was toevaluate the relationship between molecular measurements ofenterococci and E. coli and precipitation associated riverdischarge and anthropogenic variables over the past severalhundred years in sediment cores from two subwatersheds of alake system. Two locations with distinct watershed character-istics within the Lake St. Clair watershed were chosen. TheAnchor Bay watershed in northwestern Lake St. Clair washistorically agricultural lands but has transitioned to urbandevelopment. The bay sampling location is heavily influencedby water from Lake Huron and small inputs in the Anchor Baywatershed. In contrast, the mouth of the Clinton Riversampling site in western Lake St. Clair is influenced by theClinton River watershed and a steady increase in develop-ment.10 Sediment cores were taken from both locations, andassessed for vertical changes in sedimentary concentrations oftotal nitrogen, total phosphorus, and total carbon in associationto sedimentary concentrations of FIB. Additionally, the changesin historical human populations in Clinton River and Lake St.Clair watersheds, historical river discharge, and air temperaturewere evaluated against changes in FIB concentrations.

■ MATERIALS AND METHODSField Site Description, Sample Collection and Pro-

cessing. The Clinton River and Anchor Bay subwatersheds arewithin the Lake St. Clair watershed, and are 1968 and 443 km2,respectively. The sites, Anchor Bay (AB, 42°62.175′N,82°74.935′W) and Clinton River (CR, 42°59.494′N,82°79.241′W, Figure 1), were chosen because they havedistinct histories of environmental perturbation.11 Also, these

sites were reported to be sediment accumulation zones, andthus are useful sites for the reconstruction of historicalchanges.12 At each site, a pontoon boat with a PneumaticVibracore Core Sampler with 14 000 vibrations per min (GreatLakes Environmental Center, Traverse City, MI) retrieved fivesediment cores using sterile acetate butyrate tubes with aninside diameter of 9.5 cm. Latex gloves were worn whilehandling the cores and each coring tube was rinsed three timeswith 10% bleach solution and then 3 times with 10% sodiumthiosulfate prior to sediment collection. Surface sediment wasremoved with a petite ponar sampler. Surface sediments andthe cores were stored vertically on dry ice during transport, andsubsequently stored vertically at −80 °C. One frozen sedimentcore from each site was aseptically cut into 2 cm verticalsections (core length: AB = 86 cm and CR = 58 cm), andstored at −80 °C. All personnel wore clean laboratory coats,protective eye gear and changed gloves after handling eachvertical section. We removed the outer layer that was in contactwith the liner. After each core slice, the band-saw blade, movingparts, and work bench was cleaned with 70% ethanol. Therewere a total of 43 and 29 sediment sections from the AB andCR cores, respectively.

Comparison of DNA Extraction Yields and qPCRMeasurements of Enterococci 23S rDNA and E. coliuidA. A comparison of total DNA, E. coli and enterococci yieldsfrom surface sediments from Anchor Bay and Clinton River wasconducted in order to determine the optimal method for theextraction of DNA from the sediment cores. Six modificationsof three DNA extraction methods, MoBio PowerSoil (MB-PS,MoBio Laboratories, Inc., Carlsbad, CA), Mobio UltraClean(MB-UC, MoBio Laboratories, Inc., Carlsbad, CA), and amethod (EPA-DNA2) based on USEPA Method 1611,13 werecompared. Further details of the modifications and the methodsare listed in Supporting Information (SI).The AB and CR sediment sections were thawed at 4 °C

overnight and homogenized with EPA-DNA2 with G2(Geological Survey of Denmark and Greenland, GEUS,Coppenhagen, Denmark). In each section of the sedimentcores, DNA was extracted from three subsamples. A negativecontrol was analyzed for every vertical section. If amplificationwas present at <37 cT, then DNA extraction of the section wasrepeated. ENT-23 and EC-uidA were subsequently measured

Figure 1. Map of the Lake St. Clair basin, and the sediment core sites labeled with stars: Anchor Bay (AB), and the Clinton River (CR).

Environmental Science & Technology Article

DOI: 10.1021/acs.est.6b01372Environ. Sci. Technol. 2016, 50, 6902−6911

6903

Figure 2. A−H: Indictors of climate, anthropogenic attributes, and nutrient and fecal indicator bacteria loading in the Anchor Bay watershed. Theclimatic variables included (The annual average air temperature, discharge, and population were averaged within the time interval estimated in eachsediment section of the AB core): (A) air temperature at Self-ridge Air National Guard Base, Mt. Clemens, MI (1937−2012), and (B) discharge(m3/s) in the St. Clair River (1900−2012). (C) Population in the Anchor Bay watershed was estimated from census data (1900−2010). Nutrientloading in the AB core (c.1760−2012) was described with (D) total P (ug/g-dry wt), E) % total C, and F) % total N. Fecal indicator concentrationsin the AB core were measured via G) E. coli (EC), and H) Enterococci (ENT) concentrations (cell equivalents/g-dry wt, c.1760−2012). The linesrepresent the ENT and EC detection limits in the AB core sections. Data points that are not filled indicate samples that were below the detectionlimit. The error bars represent one standard error. SI Tables S2 and S3 give a further description of E. coli and enterococci concentrations in the core,respectively.

Environmental Science & Technology Article

DOI: 10.1021/acs.est.6b01372Environ. Sci. Technol. 2016, 50, 6902−6911

6904

Figure 3. A−H: Indictors of climate, anthropogenic attributes, and nutrient and fecal indicator bacteria loading in the Clinton River watershed. Theclimatic variables included (The annual average air temperature, discharge, and population were averaged within the time interval estimated in eachsediment section of the CR core): (A) air temperature at Self-ridge Air National Guard Base, Mt. Clemens, MI during 1937−2012, and (B)discharge (m3/s) in the Clinton River (1935−2012). (C) Population in the Clinton River watershed was estimated from census data (1900−2010).Nutrient loading in the CR core (c.1895−2012) was described with (D) total P (ug/g-dry wt), (E) % total C and (F) % total N. Fecal indicatorloading in the CR core was measured via (G) E. coli (EC), and (H) Enterococci (ENT) concentrations (cell equivalents/g-dry wt, c.1895−2012).The lines represent the ENT and EC detection limits in the CR core sections. Data points that are not filled indicate samples that were below thedetection limit. The error bars represent one standard error. SI Tables S2 and S3 give a further description of E. coli and enterococci concentrationsin the core, respectively.

Environmental Science & Technology Article

DOI: 10.1021/acs.est.6b01372Environ. Sci. Technol. 2016, 50, 6902−6911

6905

with qPCR as explained in SI. Recovery of ENT-23 and EC-uidA was also investigated with detailed descriptions ofmethods, standard curve performance, limits of detection, andresults in SI. Matrix interference and inhibition was evaluatedwith two replicates on every qPCR plate with a 1:1 mixture ofextracted DNA and the corresponding 104 CE/5 uL standard.Interference of >1 log difference between the 1:1 mixture andthe corresponding extracted DNA solution was not observed.There is one copy of EC-uidA per genome, while 4−6 copies

of ENT-23 per genome of enterococci.14 Therefore, one copyof EC-uidA represented one CE, whereas four copies of ENT-23 represented one CE. Based on the above conventions, theqPCR measurements of the two indicators were transformed toCE per g-dry wt. Further mention of fecal indicatorconcentrations will reference the sediment as g dried weightin order to standardize masses across the two locations andwater content of the vertical sections. Description of how watercontent was measured is described in the section below.Sediment Chronostratigraphy. Approximately, ∼30 g

(wet weight) of each sediment section was dried at 90 °C for48 h. The percent water content in each sediment section wasdetermined, and used to calculate the porosity andcommutative mass depths.15 From the measured verticalprofiles of 137Cs and excess 210Pb activity, the approximateyears of deposition were calculated as previously outlined.15

The activities of total 210Pb, 226Ra, and 137Cs were measuredusing high-resolution gamma-ray spectrometer, and the detailsare given in Jweda and Baskaran (2011). Details on the excess210Pb (210Pbxs) and 137Cs-based chronologies are discussed inBaustian et al. (in prep.). From the peak 137Cs activitycorresponding to 1963, the chronology of the sedimentarylayers was established. There was an overall agreement between210Pbxs and

137Cs based chronologies.Measurements of Sedimentary Nutrients. Concentra-

tions of total phosphorus (TP, ug/g-dry wt), percent total C (%tC), and percent total N (% tN) were measured in each driedsediment section, and are detailed in SI.Measurements of the Anthropogenic and Climate

Data. The average monthly temperature (°C) was measuredfrom Self-ridge Air National Guard Base (42°36.498′N,82°49.098′W) during 1937−2012.16 The base provides areasonable representation of air temperatures of Anchor Bayand Clinton River because of its close proximity to both sites.The monthly average discharge (m3/sec) from St. Clair River(1900−2012) was previously published17 and data iscontinuously added (http://www.glerl.noaa.gov/data/arc/hydro/mnth-hydro.html). The St. Clair River was chosenbecause it contributes the majority of the water discharged intoAnchor Bay. The daily discharge (m3/sec) of the Clinton Riverwas measured during 1935−2012 from the USGS weatherstation 04165500 (42°35.75′N, 82°54.53′W) upstream ofCR.18 The above measurements were averaged over a calendaryear.Data on human population in the communities of the

Clinton River and Anchor Bay watersheds during 1900−2010were gathered from the U.S. Census Bureau.19 Communities inthe watersheds were identified with having ≥50% of their areawithin the watershed boundaries. Based on the census data, theannual populations in the Anchor Bay and Clinton Riverwatersheds during 1900−2010 were estimated using thefollowing linear regression equations:

= − + ×P Yln( ) 38.86 0.03AB (1)

and

= − + ×P Yln( ) 53.39 0.03CR (2)

respectively, where Px was the estimated population in theClinton River or Anchor Bay watershed, and Y was the year.Each core section (2 cm vertical length) represented multiple

calendar years. Therefore, the average air temperature, riverdischarge, and estimated population represented the range ofyears encompassed in each core section.

Multiple Linear Regression Analyses. Multiple linearregression analyses evaluated two data sets: ENT concen-trations deposited during c.1932−2012 in both cores, CENT;and EC concentrations deposited during c.1932−2012 in theAB core and during c.1951−2012 in the CR core, CEC; andtheir associations to the following independent variables: EC(when CENT was the dependent variable) or ENT (when CECwas the dependent variable) concentrations, CEC or CENT,respectively; calendar year, Y; sedimentary nutrients (P, N, and,C) concentrations (ug/g-dry wt), H, N, and A, respectively;watershed population, P; river discharge, D; air temperature, T;and site, S. The ENT and EC linear regression equations were:

= × + × + ×

+ × + × + × + ×

+ × + × +

−C b C b P b Y

b H b N b A b T

b D b S b

ln( ) ln( ) P Y

H N A T

D S

ENT C EC EC

0 (3)

,and

= × + + ×

+ × + × + × + ×

+ × + × +

−C C b P b Y

b H b N b A b T

b D b S b

ln( ) b ln( ) P Y

H N A T

D S

EC C ENT ENT

0 (4)

,respectively. Also, b0 was the y-intercept, and b with a lettersubscript represented the correlation coefficient of eachindependent variable. In order to account for the time framebetween discharge and deposition into the sediments, theaverage discharge rates representative of each core section weredisplaced by three prior core sections in the analysis (e.g., themost recent sediment section represented sediment between 0and 2 cm and spanned the years 2010−2012, then the averagedischarge rate that was representative of this section wasgathered during the years 2002−2004, which was three coresections earlier between 6 and 8 cm).

■ RESULTSSedimentation Rate and Sediment Chronostratigra-

phy. Little evidence of vertical sediment mixing was observedin the 137Cs profile (SI Figure S1). The mass accumulation rates(and linear sedimentation rates) in the AB and CR cores were0.35 g/cm2 (0.39 cm/yr), and 0.42 g/cm2 (0.67 cm/yr)respectively. Therefore, the AB and CR cores representapproximately 255 and 117 years, respectively. Uniformsedimentation rate over the whole core is assumed over thistime period.

Climatic Measurements: Air Temperature and Dis-charge Rates. The average annual surface air temperatureswere averaged to represent the time interval estimated in eachsediment section, and ranged from 8.4°−10.3°, and 7.9°−10.2°C for AB and CR, respectively (Figures 2A and 3A,respectively). From 1932 to 1951 and from 1978 to 2012,the air temperature increased, while it decreased during the1951 to 1978 interval (Figures 2A and 3A).

Environmental Science & Technology Article

DOI: 10.1021/acs.est.6b01372Environ. Sci. Technol. 2016, 50, 6902−6911

6906

The annual average river discharge representative of AB andCR was averaged within the time intervals and estimated ineach sediment section. The range of discharge in the St. ClairRiver (4.52 × 103−5.84 × 103 m3/s) influencing AB was largerthan the Clinton River (0.65 × 101−3.23 × 101 m3/s, Figures2B and 3B, respectively). The discharge in St. Clair Riverdecreased from c.1971−2012 (Figure 2B). The discharge in theClinton River initially increased, then declined, and thenincreased again from 1934−1945; 1945−1963; and 1963−2012, respectively (Figure 3B).Anthropogenic Attributes: Estimated Census Popula-

tion and Nutrient Loading in the Cores. The estimatedhuman populations in the Anchor Bay and Clinton Riverwatersheds (1900−2010) were averaged to represent the timeintervals in each sediment section. In 1900 (and 2010), thepopulation densities per km2 in AB and CR watersheds were:52.76 and 51.32 people per km2, respectively (increasing to1.84 × 104 and 1.55 × 105 people per km2, respectively). TheClinton River watershed had a larger total population comparedto the Anchor Bay watershed (Figures 2C and 3C), while theestimated growth rates in both watersheds were similar. Since1900, the estimated populations in both watersheds increasedby ∼1-log (Figures 2C and 3C).The sedimentary nutrient concentrations in the CR core

steadily increased toward present day (Figure 3D−F). The TPconcentrations in the AB and CR cores were similar untilc.1966 (∼1.5 × 102 ug/g-dry wt; Figures 2D and 3D,respectively), and thereafter doubled in CR core (Figure 3D).The range of the % tC in the AB core was larger (6.2 × 10−1 −140 × 10−1 % tC) compared to the CR core (9.3 × 10−1 − 59.3× 10−1 % tC, Figures 2E and 3E, respectively), whereas therange of the % tN in the AB core was smaller (3.0 × 10−2−66 ×10−2 % tN) compared to the CR core (5.0 × 10−2−30 × 10−2

% tN, Figures 2F and 3F). Starting in c.1856, the % tC and %tN in the AB core began to increase >1-log until they reachedtheir largest values in c.1896 (140 × 10−1 % tC, and 6.6 × 10−1

% tN, respectively), and then began to decrease by >1-log untilthe concentrations returned to the pre-1856 values at c.1932(Figure 2E−F). The % tC (and % tN) was lower in the CRcore than in the AB core during two intervals: pre-1925 (pre-1918), and post-2007 (post-2009; Figure 3E−F). Theconcentrations of sedimentary nutrients measured in AB andCR are detailed in SI Tables S2−S3, respectively.

Fecal Indicator Concentrations in Sediment Cores.ENT-23 and EC-uidA concentrations were measured fromthree technical replicates in each vertical section of the AB andCR cores. In the AB (and CR) sediment sections, 9% (13%)and 72% (41%) of the samples were below the detection limitsof ENT-23 and EC-uidA, respectively, and those ENT and ECconcentrations were reported at their detection limits (Figures2G−H and 3G−H). The detection limits were dependent onthe core depth and were specific for each marker. The values ofthe detection limits of the nondetects are listed in SI Tables S2and S3. The oldest detectable ENT (and EC) concentrations inthe AB and CR cores were found in a section representingc.1760 (c.1776), and c.1911 (c.1924), respectively (Figures2G−H and 3G−H). The EC concentrations in the CR and ABcores increased toward present day in both sites, and rangedfrom 0.14 × 107 to 1.69 × 107 CE/g-dry wt, and 1.8 × 106 to8.5 × 106 CE/g-dry wt, respectively (Figures 2G and 3G).However, the ENT concentrations were at steady-state in theAB core during the following time intervals: c.1760−c.1860(∼0.01 × 106 CE/g-dry wt), and c.1910−c.2003 (∼1.0 × 106

CE/g-dry wt, Figure 2H). ENT concentrations in CR increasedtoward present day, and ranged from ∼0.03 × 105 to 9.9 × 105

CE/g-dry wt (Figure 3H). The EC and ENT concentrations inthe AB and CR cores are further detailed in Table S2 and S3,respectively.

Linear Regression Analyses. Multiple linear regressionanalyses evaluated ENT concentrations (deposited duringc.1932−2012 in both cores), and EC concentrations (depositedduring c.1932−2012 in the AB core and during c.1951−2012 inthe CR core), and their associations to the followingindependent variables: year, site, EC or ENT concentrationin the core sections (opposite of the dependent variable), riverdischarge, air temperature, sedimentary nutrient concentrations(% tC, % tN, and TP), and estimated population in thewatersheds. The R2 values (and sample size) for the ENT andEC data sets were 0.86 (n = 36), and 0.72 (n = 33), respectively(Table 1). The ENT and EC linear regression equations were

= × + × + ×

− × + × − ×+ × + × + ×−

C C P Y

H N AT D S

ln( ) 0.682 ln( ) 0 0.008

0.003 0.822 0.0160.077 0.001 3.97725.252

ENT EC

(5)

Table 1. Linear Regression Analysis Evaluated the Associations of (1) Enterococci Concentrationsa in Core Sections DepositedDuring c.1932-2012 in Both Cores, CENT; and (2) E. coli Concentrationsa Deposited in Core Sections During c.1932−2012 inthe AB Core and During c.1951−2012 in the CR core, CEC, to the Following Independent Variables: E. coli Concentrations(When CENT Was the Dependent Variable), CEC or Enterococci Concentrations (When CEC Was the Dependent Variable), CENT;Calendar Year, Y; Sedimentary Nutrient Loading (TP, % tN, and, % tC), H, N, and A, Respectively; Watershed Population inAnchor Bay and Clinton River, P; St. Clair River and Clinton River Discharge Rateb, D; Air Temperature, T; and Site, S. TheTable Outlines the Correlation Strengths (and p-Values) of the Independent Variables to the Dependent Variables, ENT andEC Concentrations. Bold Indicates Significant p-Values

correlation strength of the independent variables (p-value)

data seta

(sample size) R2 P Y H N A T D S C

CENT 0.860 0.570 0.779 0.471 0.421 0.483 0.358 0.168 −0.161 0.791c

(n = 37) (p = 0.003) (p = 0.483) (p = 0.306) (p = 0.145) (p = 0.492) (p = 0.804) (p = 0.046) (p = 0.175) (p = 0.079)

CEC 0.724 0.478 0.613 0.430 0.269 0.408 0.214 0.068b −0.046 0.715d

(n = 33) (p = 0.023) (p = 0.498) (p = 0.289) (p = 0.038) (p = 0.029) (p = 0.018) (p = 0.058) (p = 0.117) (p = 0.230)aEnterococci (CENT), and E. coli (CEC) concentrations were normalized to cell equivalents per g dry wt. bThere was a ∼6 year lag in the dischargerates in the E. coli data set. cMeasured the coefficient and significance of the independent variable, CEC.

dMeasured the coefficient and significance ofthe independent variable, CENT.

Environmental Science & Technology Article

DOI: 10.1021/acs.est.6b01372Environ. Sci. Technol. 2016, 50, 6902−6911

6907

, and

= × + × − ×

+ × − × + ×− × + × + ×

+−

C C P Y

H N AT D S

ln( ) 0.116 ln( ) 0 0.004

0.002 0.589 0.0240.349 0.001 2.309

19.466

ENTEC

3

(6)

,respectively. The ENT concentrations in both cores weresignificantly correlated to the discharge in the Clinton Riverand the St. Clair River (p = 0.046), and the estimatedpopulation in the Clinton River and Anchor Bay watersheds (p= 0.003, Table 1). The EC concentrations in both cores weresignificantly correlated with % tN and % tC (p = 0.038, and

0.029, respectively), estimated population in the Clinton Riverand Anchor Bay watersheds (p = 0.023), and air temperature (p= 0.018, Table 1). ENT and EC concentrations were notsignificantly correlated to each other (p = 0.23 and 0.079,respectively, Table 1), nor were the average FIB concentrationssignificantly different between the sites (p > 0.1, Table 1).

■ DISCUSSIONPrevious investigations have measured chemical contaminants,total P, organic C, and organic N in sediment cores in order toreconstruct large scale assessments of historical land manage-ment practices.20,21 Our study is one of the first to evaluate theassociations of anthropogenic activities, and climate variables tomolecular measurements of FIB concentrations in sediment

Figure 4. A−B: Illustration of the data patterns of climatic (discharge of A. St. Clair River and B. Clinton River) and anthropogenic (population inwatershed, and policy changes) variables measured from historical data, and nutrient (total phosphorus, % total C, and % total N) and fecal indicatorbacteria (E. coli and enterococci cell equivalents) measured in sediment cores from the (A) Anchor Bay and (B) Clinton River watersheds. TheAnchor Bay and Clinton River sediment cores spanned the years: c.1760−2012 and c.1895−2012, respectively. The NDPES is the Nationaldischarge pollutant elimination system.

Environmental Science & Technology Article

DOI: 10.1021/acs.est.6b01372Environ. Sci. Technol. 2016, 50, 6902−6911

6908

cores over 100+ years. We contrasted two locations in the LakeSt. Clair watershed: (1) near the mouth of the Clinton River(CR), and (2) Anchor Bay (AB) in northwestern Lake St. Clair.Lake St. Clair and its watershed are important to the GreatLakes basin as the key body of water connecting the Upper andLower Great Lakes, as well as home to >4 million people, and isthe drinking water source for metropolitan Detroit andsouthwestern Ontario.22

Analysis of the AB core determined that the bottom mostsections (c.1760−c.1800) were deposited when the watershedwas a trading outpost and a Native American settlement.23

During this time, % tC, % tN, and FIB measurements were atsteady state (Figures 2D−H and 4A). Small scale logging of thearea began in 1766 and timber was rafted on the St. Clair Riverto Detroit.24 Wide scale logging, timber processing, andtransport began after the Swampland Act of 1850,23 whichallowed settlers to obtain wetlands at no cost if they weredrained and developed.25 Subsequently, the Anchor Baywatershed was deforested by the early 20th Century.24 Theeffects of deforestation, destruction of wetlands, and thebooming logging industry were evident as increases in % tC,% tN, and enterococci concentrations until around the start ofthe 20th Century (Figures 2D−F and 4A). Also, the C:N ratiobefore the Swamplands Act of 1850 was ∼11:1,26 indicatingthat the tC and tN were mainly from aquatic sources. After thepolicy change, the C:N ratio increased to 21.85,26 whichindicate a strong input of tC and tN from terrestrial, likelywoody, sources.27 Similarly, a previous study also observedincreases in organic matter and C:N ratios measured insediment cores from the Great Lakes that coincided withdeforestation.28 The formation of the second steady stateconcentration of ENT at 1-log larger than the predeforestationlevels, and the return to the predeforestation levels of % tC, and% tN in the AB core began in c.1896 (Figures 2E−G and 4A).This demonstrated that activities in the watershed perhapsformed a new “equilibrium”, which could have been the resultof forest regrowth or widespread agricultural practices thatbuffered nutrient loading and wastewater/fecal pollution inputsinto Anchor Bay. Similar to our findings, concentrations of totalorganic C in a sediment core from Crawford Lake, Ontario,Canada increased then returned to predeforestation levelsobserved after two different deforestation events linked toIroquoian settlements and westernized farming (c.1268 andc.1867, respectively).29 Increases in EC and TP concentrationsin the AB core starting c.1890 and c.1849 (Figures 2D and G,and 4A), respectively, are likely related to the increase inpopulation of the watershed, which may be related towastewater production. The sedimentary % tN and % tCdeclined and remained rather stable during c.1932−c.1992,which occurred during the wide scale development ofMichigan’s shorelands25 (Figures 2E−F, and 4A). Suchdevelopment appeared to be less of an impact on organic Cand N loading than deforestation. In 1972, the NationalPollutant Discharge Elimination System (NDPES) beganregulating P loading in the Great Lakes,30 and during thisperiod, its concentration remained rather stable (Figures 2Dand 4A), which may be attributed to removal of P in wastewatertreatment plants and a ban on P in detergents31 (Figures 2Aand 4A). The ENT and EC concentrations began to decrease in2003, which could be the result of programs initiated in 2002 inSt. Clair and Macomb counties to identify and fix failing septicsystems33 (Figures 2G−H; and 4A).

The larger sedimentation rate at CR is likely due to highersediment supply (per unit volume of water discharge) from theClinton River compared to St. Clair River. Overall, the riverdischarge, FIB concentrations, % tC and % tN in the CR corehave increased since c.1895 (and since c.1934 for riverdischarge only, Figure 3B, and 3D−H). The increases of %tC, % tN, and FIB concentrations in the Clinton River werepreviously attributed to failing septic systems, stormwater, andrunoff.34 The sedimentary TP profile suggest that its loadingbegan to increase in c.1966 (Figures 3D and 4B). However, therate of tC and tN loading decreased in c.1962 (Figures 3E−F),and ENT concentrations stabilized during c.1969−c.1987(Figure 3H, and 4B), perhaps due to the elimination of rawsewage inputs and regulation of point source pollutioninstituted by National Pollutant Discharge Elimination initiatedregulation of P in 1972.35 Discharge at the mouth of theClinton River decreased in 1951 (Figures 3B, and 4B) as aresult of the construction of a spillway to prevent flooding,35

which did not curb the loading of FIB, TP, tC, or tN into thewaterway (Figures 2D−H, and 4B). The Clinton River was firstlisted as an EPA Area of Concern in 1988, and a remedialaction plan was drafted in order to initiate remediation of thewatershed. Revitalization efforts within the Clinton Riverwatershed may have decreased the C:N ratio c.1987.26

However, delisting of the Clinton River has yet to beaccomplished, and is evidenced by the increasing FIB, % tN,and % tC abundances in the core since c.1987 (Figures 3D−H,and 4B).Our study is one of the first to measure genetic markers from

ENT and EC in sediments that were deposited in the Lake St.Clair watershed >200 yrs ago. Previously, culturable andmolecular measurements of ENT and EC were enumeratedfrom 60 cm beneath the surface sediment in Lake Geneva,Switzerland.9,36 Both studies further suggest that sediments arelong-term reservoirs for FIB that originate in gastrointestinaltracts of humans and or mammals. Attachment of cells to solidparticles like sediment may protect the cell wall fromdegradation with unknown mechanisms.37 Additionally, sedi-ments may also provide organic materials and protect cells fromsunlight, and potentially predation.38 Therefore, sediments canprovide evidence of historical variation in inputs. This study isunable to ascertain the concentration of FIB at the time ofdeposition. However, a large scale bacteriological study fundedby the International Joint Commission in 1912 observedconcentrations of total coliforms from 5 to 200 CFU/100 mLin surface waters of the Great Lakes.39 Decay of the moleculartargets is a consideration and a research need. To ourknowledge there is little research that has evaluated the ratesof decay of FIB longer than one year in any matrix. Therefore,it would be inappropriate to extrapolate decay rates fromrelatively short-term studies to the longer time scales that wereevaluated in our study. The EC-uidA measurements were notsufficiently sensitive at all sediment depths and consequentlythe profile of the EC concentrations in the cores is incomplete.However, when detected, the EC concentrations were largerthan ENT (Figures 2G, and 3G), which echoes in measure-ments of culturable ENT and EC from sediment cores obtainedfrom Lake Geneva.9

Our study evaluated the statistical associations of anthro-pogenic (human population in watershed, and sedimentarytotal P, % tN, and % tC), and climate (river discharge, and airtemperature) variables to FIB concentrations in both cores.The ENT and EC concentrations in both cores were

Environmental Science & Technology Article

DOI: 10.1021/acs.est.6b01372Environ. Sci. Technol. 2016, 50, 6902−6911

6909

significantly correlated to population (p = 0.003, and 0.023,respectively). A previous study also reported that urbanizedareas of a marine estuary experienced larger fecal coliformconcentrations.40 EC concentrations were significantly asso-ciated with % tC and % tN (p = 0.029, and 0.038, respectively)in our study. Similarly, culturable EC concentrations insediments increased with organic matter and N concentra-tions.38 The air temperature was negatively associated with ECconcentrations in the cores (p = 0.018). Similarly, EC 23SrDNA demonstrated longer persistence in beef manureamended soil stored at 10 °C than at 25 °C.41 ENTconcentrations were significantly correlated to river dischargein our study (p = 0.046), suggesting that ENT attachment tosuspended particles facilitates their movement to the benthos.42

Therefore, increased discharge partially made up of wastewatereffluent, discharges from combined sewer overflow events, andexcessive nonpoint runoff from flooding could increase ENTconcentrations in surface sediments. The lack of correlationbetween ENT and EC concentrations could be due to theirdistinct rates of decay,37 resuspension and sedimentation.The findings from this study demonstrate that sediments are

reservoirs for aquatic nutrients and genetic markers thatrepresent fecal pollution Also, sediment core investigationsallow for a singular method to measure concentrations of FIB,which is not possible with sampling in the water column.Therefore, cores can be used to track the changes in abundanceof FIB using molecular measurements. Paleolimnologicalinvestigations can be used to reveal associations betweenclimate, anthropogenic attributes, regulation, and water qualityover long time scales. Therefore, sediment core studies are anapproach which allows one to analyze historical watershedhealth, and provide a better understanding of sustainablemanagement strategies that could improve water quality.Ref 32

■ ASSOCIATED CONTENT*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/acs.est.6b01372.

1. Supplemental graphics of 137Cs radio isotope activity,DNA extraction and recovery; 2. Supplemental tableswith inventories of the concentrations of nutrients andfecal indicators of each 2 cm section of both cores; 3.Methods regarding the optimization of DNA extractionprotocols, qPCR protocols of E. coli uidA and enterococci23S rDNA, DNA recovery from core sections, measure-ments of total carbon, nitrogen and phosphorus insediment sections (PDF)

■ AUTHOR INFORMATIONCorresponding Author*Phone: +1.517.432.8185; fax: +1.517.432.1699; e-mail:brooksyo@msu.edu.NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThis study was partially funded by the NSF Water Sustainabilityand Climate award #1039122 and National Oceanic andAtmospheric Administration Center of Excellence for GreatLakes and Human Health #NA04OAR4600199. The authorswould like to thank the following people for helping with this

study: Great Lakes Environmental Center, Inc., and ZacharyGeurin for their help collecting the sediment cores; Tiong GimAw, Brian Graff, and Georgia Mavrommati for help processingsediment cores; Melissa Erickson for initial analysis of theconcentrations of total C and total N; Anupam Kumar foranalysis of radioisotope activity; Hasand Gandhi for measuringand analyzing sedimentary total C and total N; Wenjuan Ma forhelp with statistical analysis; and Seth Hunt for analysis of totalP in sediment cores.

■ REFERENCES(1) Wolf, H. M. The coliform count as a measure of water quality. InWater Pollution Microbiology; Mitchell, R., Ed.; Wiley-Interscience:New York, NY, 1972.(2) USEPA. Ambient Water Quality Criteria for Bacteria - 1986;Washington, DC, 1986.(3) USEPA. Bacterial Water Quality Standards for Recreational Waters555 (Freshwater and Marine Waters) Status Report; Washington, DC,2003.(4) USEPA. Recreational Water Quality Criteria; Washington, DC,2012.(5) Eichmiller, J. J.; Hicks, R. E.; Sadowsky, M. J. Distribution ofgenetic markers of fecal pollution on a freshwater sandy shoreline inproximity to wastewater effluent. Environ. Sci. Technol. 2013, 47 (7),3395−3402.(6) Wheeler Alm, E.; Burke, J.; Spain, A. Fecal indicator bacteria areabundant in wet sand at freshwater beaches. Water Res. 2003, 37 (16),3978−3982.(7) Pote, J.; Haller, L.; Kottelat, R.; Sastre, V.; Arpagaus, P.; Wildi, W.Persistence and growth of faecal culturable bacterial indicators in watercolumn and sediments of Vidy Bay, Lake Geneva, Switzerland. J.Environ. Sci. (Beijing, China) 2009, 21 (1), 62−69.(8) Craig, D. L.; Fallowfield, H. J.; Cromar, N. J. Use of microcosmsto determine persistence of Escherichia coli in recreational coastalwater and sediment and validation with in situ measurements. J. Appl.Microbiol. 2004, 96 (5), 922−930.(9) Thevenon, F.; Regier, N.; Benagli, C.; Tonolla, M.; Adatte, T.;Wildi, W.; Pote,́ J. Characterization of fecal indicator bacteria insediments cores from the largest freshwater lake of Western Europe(Lake Geneva, Switzerland). Ecotoxicol. Environ. Saf. 2012, 78, 50−56.(10) Fry, J. A.; Xian, G.; Jin, S.; Dewitz, J. A.; Homer, C. G.; Yang, L.;Barnes, C. A.; Herold, N. D.; Wickham, J. D. Completion of the 2006National Land Cover Database for the conterminous United States.Photogramm. Eng. Remote Sensing 2011, 77, 858−566.(11) Healy, D. F.; Chambers, D. B.; Rachol, C. M.; Jodoin, R. S.Water quality of the St. Clair River, Lake St. Clair, and their U.S.tributaries, 1946−2005: U.S. Geological Survey Scientific InvestigationsReport 2007−5172; Reston, VA, 2008.(12) International Joint Commission; Upper Great Lakes Connect-ing Channels Study. Upper Great Lakes Connecting Channels Study:Volume II- Final Report; 1988.(13) USEPA. Method 1611: Enterococci in Water by TaqManQuantitative Polymerase Chain Reaction (qPCR) Assay; Washington,DC, 2012.(14) USEPA. Method A: Enterococci in Water by TaqMan ®Quantitative Polymerase Chain Reaction (qPCR) Assay; Washington,DC, 2010.(15) Jweda, J.; Baskaran, M. Interconnected riverine−lacustrinesystems as sedimentary repositories: Case study in southeast Michiganusing 210Pb and 137Cs-based sediment accumulation and mixingmodels. J. Great Lakes Res. 2011, 37 (3), 432−446.(16) Midwestern Regional Climate Center. Cli-MATE: MRCCApplication Tools Environment http://mrcc.isws.illinois.edu/CLIMATE/Station/Annual/StnAnnualBTD.jsp.(17) Hunter, T. S.; Croley, T. E. I. Great Lakes Monthly HyrdologicData, NOAA Data Report ERL GLERL; Springfiled, Virginia, 1993.(18) USGS. USGS Surface-Water Annual Statistics for the Nationhttp://waterdata.usgs.gov/nwis/annual?referred_module=sw&site_

Environmental Science & Technology Article

DOI: 10.1021/acs.est.6b01372Environ. Sci. Technol. 2016, 50, 6902−6911

6910

no=04165500&por_04165500_2=892169,00060,2,1934,2014&year_type=W&format=html_table&date_format=YYYY-MM-DD&rdb_compression=file&submitted_form=parameter_selection_list.(19) SEMCOG. Historical Population and Employment by Minor CivilDivision; Detroit, MI, 2002.(20) Furl, C. V.; Meredith, C. A. Mercury accumulation in sedimentcores from three Washington state lakes: evidence for local depositionfrom a coal-fired power plant. Arch. Environ. Contam. Toxicol. 2011, 60(1), 26−33.(21) Kaushal, S.; Binford, M. W. Relationship between C: N ratios oflake sediments, organic matter sources, and historical deforestation inLake Pleasant, Massachusetts, USA. J. Paleolimnol. 1999, 22 (4), 439−442.(22) Baustian, M. M.; Mavrommati, G.; Dreelin, E. A.; Esselman, P.;Schultze, S. R.; Qian, L.; Aw, T. G.; Luo, L.; Rose, J. B. A one hundredyear review of the socioeconomic and ecological systems of Lake St.Clair, North America. J. Great Lakes Res. 2014, 40 (1), 15−26.(23) Unknown. History of St. Clair County, Michigan, Containing anAccount of Its Settlement, Growth, Development and Resources, Its WarRecord, Biographical Sketches, The Whole Preceded by a History ofMichigan; A.T. Andreas and Co.: Chicago, IL, 1883.(24) Crampton, E. J. History of the St. Clair River; St. ClairRepublican: St. Clair, MI, 1921.(25) Fishbeck, Thompson, Carr & Huber, I. Anchor Bay WatershedPlan 2006; Grand Rapids, 2007.(26) Baustian, M. M.; Baskaran, M. M.; Brooks, Y. M.; Leavitt, P. R.;Liu, B.; Ostrom, N.; Stevenson, R. J.; Rose, J. B. Paleo-environmentalexamination of human-driven ecosystem change in Lake St. Clairregion of Laurentian Great Lakes basin. in prep.(27) Meyers, P. A.; Ishiwatari, R. Lacustrine organic geochemistryan overview of indicators of organic matter sources and diagenesis inlake sediments. Org. Geochem. 1993, 20 (7), 867−900.(28) Meyers, P. A. The Interactions Between Sediments and Water;Kronvang, B., Faganeli, J., Ogrinc, N., Eds.; Springer Netherlands:Dordrecht, 2007.(29) Ekdahl, E. J.; Teranes, J. L.; Guilderson, T. P.; Turton, C. L.;McAndrews, J. H.; Wittkop, C. A.; Stoermer, E. F. Prehistorical recordof cultural eutrophication from Crawford Lake, Canada. Geology 2004,32 (9), 745.(30) Harrington-Hughes, K. Great Lakes Water Quality: A ProgressReport. J. (Water Pollut. Control Fed. 1978, 50 (8), 1886−1888.(31) International Joint Commission. Report on Great Lakes WaterQuality for 1972; Windsor, Canada, 1973.(32) Associated Press. Report: Dredging Water Causes Huge GreatLakes Water Loss. MLive Media Group. Traverse City, MI,August 14,2007.(33) St. Clair County Health Department. An Evaluation of AnchorBay Watershed Planning Activities and Monitoring Data; Port Huron,MI, 2009.(34) Environmental Conuslting and Technology, I. Water QualitySampling and Analysis Final Report; Clinton Township, MI, 2007.(35) Michigan Department of Environmental Quality. Clinton RiverRemedial Action Plan; Lansing, MI, 1988.(36) Devarajan, N.; Laffite, A.; Graham, N. D.; Meijer, M.; Prabakar,K.; Mubedi, J. I.; Elongo, V.; Mpiana, P. T.; Ibelings, B. W.; Wildi, W.;et al. Accumulation of Clinically Relevant Antibiotic-Resistance Genes,Bacterial Load, and Metals in Freshwater Lake Sediments in CentralEurope. Environ. Sci. Technol. 2015, 49 (11), 6528−6537.(37) Brooks, Y.; Aslan, A.; Tamrakar, S.; Murali, B.; Mitchell, J.; Rose,J. B. Analysis of the persistence of enteric markers in sewage pollutedwater on a solid matrix and in liquid suspension. Water Res. 2015, 76,201−212.(38) Haller, L.; Amedegnato, E.; Pote,́ J.; Wildi, W. Influence ofFreshwater Sediment Characteristics on Persistence of Fecal IndicatorBacteria. Water, Air, Soil Pollut. 2009, 203 (1−4), 217−227.(39) Rose, J.; Dreelin, E. Effective Cross-Border Monitoring Systemsfor Waterborne Microbial Pathogens. Water Intell. Online 2008, 7,1476−1777.

(40) Kelsey, H.; Porter, D. E.; Scott, G.; Neet, M.; White, D. Usinggeographic information systems and regression analysis to evaluaterelationships between land use and fecal coliform bacterial pollution. J.Exp. Mar. Biol. Ecol. 2004, 298 (2), 197−209.(41) Rogers, S. W.; Donnelly, M.; Peed, L.; Kelty, C. a; Mondal, S.;Zhong, Z.; Shanks, O. C. Decay of bacterial pathogens, fecal indicators,and real-time quantitative PCR genetic markers in manure-amendedsoils. Appl. Environ. Microbiol. 2011, 77 (14), 4839−4848.(42) Jeng, H.; England, A.; Bradford, H. Indicator OrganismsAssociated with Stormwater Suspended Particles and EstuarineSediment. J. Environ. Sci. Health, Part A: Toxic/Hazard. Subst. Environ.Eng. 2005, 40 (4), 779−791.

Environmental Science & Technology Article

DOI: 10.1021/acs.est.6b01372Environ. Sci. Technol. 2016, 50, 6902−6911

6911